[tei-council] teilite.dtd standin

[Lou Burnard]

I had a vague memory that we considered but decided against doing a 
redirection though I can't remember why. But if it's possible, I agree 
this would be preferable.



On 27/10/12 12:08, Sebastian Rahtz wrote:
> On 27 Oct 2012, at 09:54, Lou Burnard <lou.burnard at retired.ox.ac.uk> wrote:
>
>> On 26/10/12 22:36, Sebastian Rahtz wrote:
>>> sorry to be a party pooper, but this is (IMHO) an awful idea.
>> so what would you propose we do instead?
>>
> My preference is for a rewrite teilite.{dtd,rnc,rng) -> tei_lite.{dtd,rnc,rng)
> because for most people it will just work. My second preference
> is for no file at all, and trigger a comprehensible 404 error.
>
>>>   My
>>> file containing
>>>
>>>   <!DOCTYPE ...." SYSTEM "...tei_lite.dtd">
>>>
>>> will now fail catastrophically, saying the DTD is invalid,
>>> and I will have no idea why.
>> Possibly you'll think, hmm, maybe I should take a look at that file?
>>
> I fear that we come from a generation which thinks like that. I have
> no confidence many other people conceptualize the world like that.
>
> Consider the person with a repository of 1000 files, in P5, with
> a  DOCTYPE pointing at Lite from the web site.  Possibly
> these links are auto-generated. The user sees HTML generated on
> the fly. Suddenly the HTML stops working. The error, when some
> poor sys person follows up, says that the DTD is invalid. They follow
> up, and find that the document calling itself teilite.dtd isn't a DTD at all!
> I call that counter-intuitive. If the error is a 404, they say "oh well, links
> break on the web, I wonder where its moved to" and go look.
>
> You may well reply that this scenario doesnt exist ....
>>> Making the file non-existent
>>> would be a much cleaner error.
>> Is that your proposal? we simply remove the file teilite.dtd wherever we can? I dont mind doing that, but it doesnt help with the fact that lots of people point to copies of it we don't control.
> thats an issue, always has been, i agree
>
>>   If the one we do control explains itself, maybe that's a good thing?
>>
> sure, its an argument.
>
> would you like it if, when you typed "perl", you got a message
> saying that the file /usr/bin/perl was not executable, and you found,
> on examining it, that it was in fact a README saying "gone fishing, use python instead"?
> of course, you'd prefer it if that /usr/bin/perl _was_ executable, and gave you
> a human readable message when you ran it. But we cant do that, since DTDs
> dont do anything.  But having a file which claims (by its suffix and by its location)
> to be a DTD, but isnt one, seems wrong.
> --
> Sebastian Rahtz
> Director (Research Support) of Academic IT Services
> University of Oxford IT Services
> 13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431
>